Objective: To examine cognitive predictors of functional performance and dexterity using an advanced upper-limb prosthetic device. Method: Veterans (n = 14) with upper-limb amputation (transradial, transhumeral, or full-arm) underwent tests of neuropsychological functioning and dexterity (Jebsen-Taylor Hand Function Test [JTHF]), and a clinician-based functional rating (Activities Measure for Upper-Limb Amputees [AM-ULA]) using the prosthetic device at study baseline. Prosthetic measures were repeated after a variable in-laboratory prosthetic training period. Results: Multiple linear regression showed Wechsler Adult Intelligence Scale-IV Digit Span (DS) and oral Trails B (TMT-B) performance explained 54% of the variance in JTHF light items per second (R2 = .54, F(2, 10) = 5.89, p < .02) and 40% of the variance in JTHF heavy items per second (R2 = .38, F(2.10) = 3.17, p < .09). DS, TMT-B, and Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) Story Recall explained 77% of the variance in AM-ULA (R2 = .77, F(3, 9) = 10.07, p < .00). Cognitive predictors remained significant after accounting for age and education. Dexterity performance was significantly better in the transradial group versus all others for JTHF light and heavy items. Conclusion(s): Measures of attention and working memory significantly predict post-training performance on tests of dexterity and functional ability using an advanced upper-limb prosthetic device. Results underscore the predictive value of neuropsychological assessment for short-term prosthetic proficiency, allowing improved individualized training. Ongoing work on this project will examine cognitive predictors of long-term success.